BERNARD NIHAL PRAME * AND JANAKA AJITH PREMA. Geological Survey and Mines Bureau, 569, Epitamulla Road, Pitakotte, Sri Lanka ABSTRACT INTRODUCTION

Transcription

1 PETROLOGY OF CHARNOCKITIC GRANULITES AND CALC-SILICATE GRANULITES FROM SOUTH-SOUTHEASTERN HIGHLAND COMPLEX OF SRI LANKA: FURTHER CONSTRAINTS FOR PHYSICO-CHEMICAL CONDITIONS OF THEIR METAMORPHIC EVOLUTION BERNARD NIHAL PRAME * AND JANAKA AJITH PREMA Geological Survey and Mines Bureau, 569, Epitamulla Road, Pitakotte, Sri Lanka * Corresponding Author: -bernardprame@yahoo.com ABSTRACT In southern and southeastern parts of the Highland Complex (HC) of Sri Lanka, garnetiferous charnockitic granulites of granitic-granodioritic and tonalitictrondjhemitic compositions are ubiquitous. Minor amounts of calc-silicate rocks and marble are intercalated with these rocks particularly in the southeastern HC. These two lithologies provide unique opportunity to study the physical conditions and fluid regime of their granulite facies metamorphism in two different chemical systems. Mineral chemistry and petrology of thirty (30) charnockitic granulites and five (05) calc-silicate rocks were studied to constrain the P-T conditions of granulite facies metamorphism and understand the characteristics of fluid regime. Phase equilibria in calc-silicate rocks clearly indicate that peak metamorphic temperatures were not less than 875 o C. In contrast, garnet-orthopyroxene and garnetclinopyroxene thermometry of charnockitic granulites yields a wide range of temperatures from 650 o C to 900 o C indicating gross re-setting of cations in some of the samples. Garnet-pyroxene geobarometry yields paleopressure estimates from about 7.5 kb in the southwestern (Matara) area to over 10 Kb in the boundary area of southeastern HC including Kataragama area. Thus, in conformity with previous studies in other parts of the HC there is a pressure gradient of about 3 Kb from western HC to eastern HC in the studied area. In Mg-depleted charnockitic rocks, garnet coronas are formed at the expense of pyroxene and plagioclase while in calc-silicate rocks secondary garnet is formed at the expense of scapolite + wollastonite or wollastonite + plagioclase. Formation of secondary garnet in charnockitic granulites and calc-granulites as a mineral reaction product and exsolution of extremely iron-rich orthopyroxene (Fs 95 ) in Mg-depleted charnockitic rocks can be attributed to an early phase of near isobaric cooling. Mineral paragenesis and T-CO 2 diagrams imply that calc-silicate rocks from different localities had varying CO 2 activities (X CO2 from 0.2 to 0.5). One calc-silicate rock indicates even higher CO 2 activity. Oxygen fugacity calculations for charnockitic granulites also imply two different ranges of oxygen fugacity, rocks of granitic-granodioritic composition having relatively low f O2 values compared to those of tonalitictrondjhemitic composition. This could be the result of marked difference of bulk X Fe values in these two rock types. Varying X CO2 values in calc-granulites and contrasting f O2 values in two different pyroxene granulite types can be best explained by internal fluid buffering and therefore these findings are at odds with hypothesis that advocates pervasive fluid infiltration during granulite facies metamorphism. Keywords: Mineral reactions, Geothemo-barometry, P-T path, Highland Complex INTRODUCTION Petrological studies of granulites may reveal valuable information on processes and physicochemical conditions prevailed in the lower crust. Two highly useful lithologies in this context are garnet-pyroxene granulites and calc-granulites. Studies in which garnet-pyroxene-plagioclasequartz assemblage has been employed to decipher P and T conditions and P-T-t path of 171

2 ancient lower crustal rocks are numerous (Hormann et al., 1980; Raith et al., 1983; Newton, 1983; Perkins and Chipera, 1985; Schumacher et al., 1990; Faulhaber and Raith, 1991). Though limited in numbers, studies based on calc-silicate assemblages are also on increase (Schenk, 1984; Warren et al., 1987; Harley and Buick, 1992; Harley et al., 1994; Bhowmik et al., 1995; Shaw and Arima, 1996; Sengupta et al., 1997). When these two types of lithologies are intercalated in a terrain, the possibility of constraining the peak P-T conditions, retrograde evolution and characterization of fluid regime are greatly enhanced as two independent sources of P-Tfluid information become available. In southern and southeastern areas of the HC of Sri Lanka garnetiferous charnockitic pyroxene granulites are ubiquitous. Geological mapping has revealed that there are sporadically distributed thin marble bands and lenses or boudins of calc-silicate rocks intercalated with pyroxene granulites, particularly near the boundary between the Highland and Vijayan Complexes (Figure 1). These occurrences offer an opportunity to characterize the physicochemical conditions of metamorphism using phase equilibria in two different chemical systems namely, Na 2 O CaO K 2 O FeO MgO Al 2 O 3 SiO 2 and CaO-MgO-FeO-Al 2 O 3 -SiO2. The present study-area (Figure 1) was selected mainly because of this reason, also considering the fact that the previous geothermobarometric surveys have largely excluded this part of the HC (Faulhaber and Raith, 1991; Schumacher and Faulhaber, 1994). As such, thirty (30) charnockitic pyroxene granulite samples, and five (05) calc-silicate rock samples from the southeastern HC of Sri Lanka were studied in order to understand P, T and fluid conditions of granulite facies metamorphism and characteristics of the P-T-t path of this part of the crust. Previous studies based on cation exchange thermometry of Grt-Px and Grt-Crd have produced a wide range of metamorphic temperatures for the HC. Therefore, one of the main objectives of this work was to bracket the peak metamorphic temperature of these rocks as Fig. 1 Geological map of Southern and Southeastern Sri Lanka showing sample distribution 172

3 narrowly as possible. GEOLOGICAL SETTING AND FIELD RELAT- IONS Basement rocks of Sri Lanka are divided into three major lithotectonic units, namely the HC, the Vijayan Complex (VC) and the Wanni Complex (WC) ( Milisenda et al., 1988; Cooray, 1994). The HC is mainly made up of highly metamorphosed supracrustal rocks and orthogneisses including charnockitic gneisses, charnockites and metabasites of which the Ndmodel ages range from 2 to 3 Ga. (Kröner et al., 1991). The VC on right is a younger Nd-model age province (1-2 Ga) made up of hornblendebiotite gneisses and migmatitic gneisses, tectonically juxtaposed to the older HC (Kröner et al., 1991). WC on left is also a younger Ndmodel age province but its lithology differs from VC in having some pelitic rocks and charnockitic rocks. Study-area mostly covers the southernmost part of the HC dominated by granulite facies rocks. A few samples are from the granulite klippen within the amphibolite facies VC located east of the HC. Main rock types exposed in the area are charnockitic granitoids (pyroxene granulites) and garnetiferous quartzo-feldspathic rocks. Metabasite, metapelite and marble occur as minor constituents. Charnockitic rocks in the central HC are closely associated with typical metasedimentary marker beds such as quartzite, marble and garnet-sillimanite-graphite schist (khondalite). A marked difference between the central HC and the studied area is scarcity of such thick quartzite, marble and khondalite bands. Regional thermobarometric surveys have recorded higher paleo pressure estimates (~9 kb) in the eastern HC and lower values (~5 kb) in the westernmost parts (Faulharber and Raith, 1991; Schumacher et al., 1992; Prame, 1991). However, only a few samples from the southernmost part of the HC have been studied on previous occasions. Geochemical studies show that most of the charnockitic granitoids in the area are of igneous origin (Prame, 1997). These granitoids now occurring as highly deformed bodies or layers were emplaced into the associated supracrustal sequence at various stages of the tectonic evolution, mostly between Ma ago (Kröner and Williams, 1993; Hölzl et al., 1994). The original intrusive contacts were obscured by intense non-coaxial deformation associated with Pan-African metamorphism postulated by severe Pb-loss at some Ma ago (Kröner et al., 1994). Though detailed structural investigations have not yet been carried out in this area, field observations indicate that much of the structural interpretations employed to the central Highland Complex is valid in the study-area as well. Extreme stretching and isoclinal folding are the macro-scale evidence for earliest deformation. These isoclinal folds are re-folded into open folds. In fact, a system of early flat-lying isoclinal folds subsequently refolded into open to tight folds, is the most conspicuous structural feature in the central parts of the HC (Berger and Jayasinghe, 1976; Voll and Kleinschrodt, 1991). As shown in Figure 1, regional structural trends are mainly in NW-SE and WNW-ESE directions. Charnockitic granitoids (pyroxene granulites) of the present study consist of two closely associated groups both occurring as discrete layers or deformed bodies (Prame, 1997). Rocks of granitic-adamellitic composition are commoner than tonalitic-trondjhemitic rocks which occur in subordinate amounts (Geological Map of Sri Lanka, 1982). These two groups of rocks are not distinguished in the existing geological maps as granitoids of different chemical affinities have been mapped as charnockite or charnockitic rocks. Prame, (1997) has shown that these charnockitic rocks of granitic and tonalitic-trondjhemitic composition have the chemical characteristics of A-type and I-type granitoids respectively. It appears that relatively minor layers of metatonalitic rocks are tectonically intercalated with layers or deformed bodies of granitic composition. Field evidences for relative age relations (intrusive) between these two groups have completely been obscured by strong flattening and stretching. Wollastonite-scapolite 173

4 bearing rocks have been reported from southwestern Sri Lanka where paleopressures around 6 kb have prevailed (Hapuarachchi, 1967; Hoffbauer and Spiering, 1994). During subsequent geological mapping of the southern and southeastern parts of the HC, a number of minor calc-silicate occurrences containing grossular, scapolite, diopside, wollastonite and anorthite have been noted (Geological sheet nos. 20 and 21, Geological Survey and Mines Bureau, Sri Lanka,). Wollastonite-scapolitegarnet granulites constitute only a very minor proportion (<1%) so that they cannot be shown as separate entities in the geological maps. These minor rocks have been overlooked or mapped as impure marble/calc-silicate rocks as they often occur as lenses, minor layers or boudins within marble (crystalline limestone). A study by Mathavan and Fernando (2001) first described the mineral reactions involving grossular, wollastonite and scapolite from this area. METHOD OF STUDY AND ANALYTICAL TECHNIQUES Thirty (30) polished thin sections were prepared from pyroxene-granulite samples collected from the areas that have been excluded in previous studies. Five (05) polished thin sections representative of five calc-silicate rocks intercalated with these rocks were also prepared. Most of the mineral analyses were performed on a Cameca Camebax (WDS) electron microprobe at the Mineralogical and Petrological Institute, Bonn University. The typical operating conditions were at15 KV and 15 na probe current. Concentrations were calculated according to Pouchou and Pichoir (1984). A few analyses (sample U-143 and AJ-1) were obtained using JEOL Superprobe at the Chiba University. RESULTS AND DISCUSSION PYROXENE GRANULITES PETROGRAPHY Major mineral constituents of the studied pyroxene granulites samples are given in the Appendix Table 1. Common accessory minerals are ilmenite, apatite and zircon. Graphite, magnetite, sulphide minerals and alanite were observed in a few samples. Mineral abbreviations of Whitney and Evans (2010) were adopted whenever possible. Following peak metamorphic assemblages were identified: Grt+Opx+Cpx+Pl+Kfs+Qtz Hbl Bt Grt+Opx+Pl+Kfs+Qtz Hbl Grt+Cpx+Pl+Kfs+Qtz Hbl Bt Opx+Cpx+Pl+Kfs+Qtz Hbl Bt Cpx+Pl+Kfs+Qtz Hbl Bt The above assemblages were encountered throughout the study-area typifying the granulite facies metamorphism. Textural features such as dimensional preferred orientations displayed by pyroxene and primary hornblende, stretched quartz, fractured and dislocated garnet, deformed twin lamellae in plagioclase indicate that these rocks have undergone penetrative (grain-scale) deformation. Garnet porphyroblasts and pyroxenes are set in a quartzo-feldspathic matrix. Clinopyroxene and plagioclase sometimes occur as inclusions of garnet. The granitic rocks from the Kataragama and Kudaoya klippen (B-25, B-117, 91-14) contain extremely stretched quartz ribbons occurring within a recrystallized polygonal feldspathic matrix exhibiting typical granulitic texture (Figure 2a; micromylonite?). Feldspar grains enclosed by quartz remains extremely stretched while matrix-feldspars have recrystallized into a granoblastic-polygonal texture. Sub-grain development of these stretched quartz is common. K-feldspar commonly occurs as microperthite but it occurs in the form of microcline in some retrogressed samples collected in the vicinity of HC/VC boundary area (91-15, 91-14, 90-46). K- feldspar is scarce in K- depleted tonalitic and trondjhemitic rocks. Though most garnet and orthopyroxene represent peak granulite facies metamorphism few samples contain garnet and orthopyroxene formed after peak conditions. In magnesium depleted (Fe/Fe+Mg) =

5 mol.) granitic-adamelletic samples (90-66, B- 117, 90-25) orthopyroxene is of secondary origin while orthopyroxene is totally absent in samples that are extremely depleted in Mg (bulk Fe/Fe+Mg = 0.98 mol.; sample 91-14). MINERAL REACTIONS Two types of mineral reactions were observed in these pyroxene granulites. One reaction type produces secondary garnet at the expense of as a reactant is not explicit. Perhaps, this mineral has already been consumed or microperthite has acted as a source of Al and Ca. In fact, microperthite with corroded margins supports the latter contention. Therefore following model reactions can be proposed as the mechanism of garnet formation: clinopyroxene + microperthite (An component) garnet + quartz + opaques (1) Fig. 2 (a) Extremely stretched quartz ribbons in a recrystallized polygonal feldspathic matrix (91-14), (b) Clinopyroxene grains surrounded by atolls of secondary garnet (B-117), Note that a part of postdeformational garnet ring runs perpendicular to the direction of preferred orientation of Cpx, (c) Garnet+quartz intergrowth associated with orthopyroxene and (d) Garnet porphyroblast (+ quartz) breaking down to form orthopyroxene and plagioclase (90-34). pyroxene and feldspar, while the other type indicates instability of garnet forming orthopyroxene and plagioclase bearing retrograde assemblages. The first type is particularly common in the Mgdepleted rocks of granitic composition occurring in the easternmost area. For example, high X Fe granitic and adamellatic rocks Cpx grains are surrounded by atolls of secondary garnets, quartz and microperthite (Figure 2b). Plagioclase is rare in these rocks and therefore role of plagioclase in garnet producing reactions clinopyroxene + plagioclase (An component) garnet + quartz + opaques (2) In the mm scale, garnet forms as rings around clinopyroxene cross-cutting the main L/S fabric. Formation of such retrograde garnet is usually attributed to an episode of near-isobaric cooling following peak metamorphism (e.g. Prame, 1991). In these high X Fe samples orthopyroxene also being a retrograde mineral does not apparently contribute in garnet-forming reactions. As mentioned earlier, iron-rich orthopyroxene was initially an exsolved product 175

6 and then grew with further cooling. It is likely that they grew beyond the confines of the host Cpx after formation of garnet. In sample number 90-58, intergrowths of garnet and quartz replace orthopyroxene (Figure 2c) suggesting the reaction: orthopyroxene + plagioclase garnet + quartz (3) Garnet and quartz intergrowths are aligned parallel to the preferred orientation of orthopyroxene however, they are clearly postdeformational. Garnet pophyroblasts in samples and from the western part of the study-area breakdown to produce plagioclase and orthopyroxene (Figure. 2d) suggesting the reverse relationship; garnet + quartz orthopyroxene + plagioclase (4) Orthopyroxene partially mantled or replaced by hornblende or biotite (90-76, 91-20, 90-59) indicates hydration at a later stage. PYROXENE EXSOLUTION AND FORMATION OF IRON RICH ORTHOPYROXENE Clinopyroxene from granitic rocks and adamellitic rocks (90-66, B-117, B-25) with extremely high Fe/Fe+Mg mol. Ratios ( ) display exsolution features that could probably explain the genesis of iron-rich orthopyroxene occurring in these rocks. Several grains of these pyroxenes were examined with polarizing microscope and back-scattered electron images to understand the process of exsolution. There are two optically discernible lamellae sets (Figures 3a, 3b). A set of very thin lamellae (thickness not more than few microns; from lower right to upper left) parallel to 001, at regular intervals is most conspicuous. The other set consists of slightly thicker (10-20 m) lamellae (parallel 100 ; from lower left to upper right) of iron rich Opx (Fs 94 ), spaced farther apart. These Opx lamellae have subsequently grown as rods (Figure 3a) beyond the confines of host Cpx. The apparent angle between the two sets is about 110 o. Electron microprobe analyses on the thin 001 lamellae are inconsistent having the compositions similar to those on host Cpx (Wo 44 Fs 52 En 4 ) as well as pigeonite compositions. This may be due to the very fine nature of the lamellae. These 001 lamellae which were originally pigeonite but may have subsequently exsolved to Opx and Cpx. In Figure 3b, these lamellae patterns are depicted in a back scattered electron image. The grey coloured background is Cpx host. From upper left to lower right are thin 001 of which thickness does not exceed 2 m. There is a segmentation of these lamellae by dark strips oriented parallel to 100 Opx lamellae (from lower left to upper right). Dark segments within 001 lamellae may represent fine exsolution lamellae or a set of dislocations/cavities. Also notable is the faulting of these 001 lamellae before the formation of iron-rich Opx lamellae. The Opx lamellae within Cpx should be parallel to 100 since this is the only plane of exact dimensional fit for both ortho- and clinopyroxene (Robinson, 1982). These microtextural features suggest that the exsolution process included following steps as proposed by Ollila et al. (1988); 1. Crystallization of homogeneous Cpx, 2. Exsolution of fine pigeonite lamellae parallel to 001, 3. faulting of pigeonite lamella and host Cpx along 100 and 4. formation of iron-rich Opx along the 100 with further cooling. Iron-rich orthopyroxene (Fs 94 ) now occurring as rods and laths associated with Cpx may have been initiated in this manner and grown with further cooling. High temperature igneous lamellae are usually larger in size and fewer in number (Robinson, 1982). At low temperatures, a large number of lamellae may be initiated but because of slow diffusion rates they may only grow to very small sizes. Rather closely-spaced and very thin lamellae seen in these samples may have therefore exsolved below the peak temperature of granulite metamorphism. According to Lindsley (1983), pigeonite with X Fe around 0.80 would be stable around 825 o C at 9 Kb. However, formation of orthopyroxene lamellae and their growth may have taken place at well 176

7 below this temperature. Compositions of ironrich orthopyroxene (Fs 94 Wo 3 En 3 ) and host Cpx give temperatures as low as 600 o C (Lindsley, 1983) indicating advanced re-equilibration of cations between Cpx host and Opx that was in constant contact. However, temperature estimates obtained by employing the Grt-Opx thermometry using compositions of these Opx and secondary garnet in contact with each other are 640 o C (using Harley, 1984) and 705 o C (using Sen and Bhattacharya, 1984). Two pyroxene thermometers of Wells (1977) and Wood and Banno, (1973) yields temperatures around 775 o C. These results indicate a minimum temperature of about 725 o C for the formation of iron-rich orthopyroxene and at 700 o C, Fs 94 wouldn t be stable below a pressure of 8 kb (Bohlen and Boettcher, 1981). The ideal conditions for stabilization and growth of this ferrosilite is substantial cooling at high pressure (isobaric cooling). In sample 90-66, isolated Cpx grains are absent. Cpx occurs as inclusions within iron-rich (Fs 87 ) Opx (Figure 3c). Cpx inclusions are in fact remnants of earlier Cpx host from which Opx have been exsolved. SEM image shows the extremely fine lamellae within Cpx and Opx (Figure 3d). Within these Cpx inclusions are lamellae which occasionally give sub-calcic compositions (Wo 33 Fe 58 En 9 ). This again indicates formation of pigeonite at an intermediate stage of pyroxene exsolution. A notable feature of the high-x Fe rocks is the lack of primary Opx. This was most probably due to inadequate pressure condition to stabilize ironrich Opx at peak metamorphic temperature. However, with subsequent cooling, favourable conditions for the formation of Opx may have prevailed. GEOTHERMOBAROMETRY Petrography and mineral chemistry of garnetpyroxene granulites strongly indicate attainment of textural and chemical equilibrium during granulite facies metamorphism. However, as a response to subsequent cooling and unloading, inter-granular cation exchange and solid-solid mineral reactions took place in some local domains destroying the peak metamorphic equilibrium. Nevertheless, compositions of mineral porphyroblasts from unaffected samples or domains are most likely to represent the peak metamorphic conditions while the arrested mineral reactions and re-equilibrated domains may be useful to decipher the retrograde P-T trajectory. Ubiquitous garnet + orthopyroxene clinopyroxene + plagioclase + quartz assemblage allow the application of a number of geothermobarometers to the studied samples. For some calibrations involving pyroxenes, it was necessary to assign Fe +2 and Mg +2 to the M 3 and M 2 sites. Tetrahedral Al was taken as 2-Si and the remainder was assigned to M 3 (octahedral) site. Fe +3 was then calculated by charge balancing. M 3 and M 2 positions occupied by Fe +2 and Mg +2 were determined assigning Al +3, Cr +3, Ti +4, Fe +3 to M 3 site and Ca +2, Na +,Mn +2 to M 2 site. Then the available Fe +2 and Mg +2 were proportionally allocated to these two sites (Wood and Banno, 1973). GEOTHERMOMETRY Two geothermometers (Wood and Banno, 1973) based on Fe +2 and Mg +2 distribution between orthopyroxene and clinopyroxene and Lindsley (1983) were applied to 12 co-existing Opx-Cpx pairs at a nominal pressure of 8 kb (Appendix Table 7). The lowest temperatures; 688 o C (Wood and Banno, 1973); 630 o C (Lindsley,1983) are for sample in which Cpx occurs as inclusions within the Opx. The low temperature may be due to advanced reequilibration between Cpx inclusions and host Cpx. Temperatures for other samples vary from 742 o C to 841 o C (Wood and Banno, 1973) and from 670 o C to 780 o C (Lindsley,1983). Temperature estimates from Grt-Opx (Harley, 1984; Sen and Bhattacharya, 1984) and Grt-Cpx (Ellis and Green, 1979) pairs are also presented in Appendix Table 7. Problems associated with the composition-activity relations of garnet introduce additional uncertainties in geothermometers involving with garnet and this has been clearly demonstrated by Faulhaber and Raith (1991) using a comprehensive set of Sri Lankan samples. The calibration of Harley 177

8 (1984) yields temperatures ranging from 638 o C to 803 o C which are generally lower than those obtained from Sen and Bhattacharya (1984) by about 100 o C. Samples and which give lower temperatures (640 o C and 700 o C obtained from various calibrations scatter over a range between 662 o C and 922 o C (excluding retrogressed samples and samples with secondary Opx). There is no systematic Fig. 3 (a) Exsolution lamellae of iron-rich clinopyroxene from granitoid rocks near Kataragama (B- 25), (b) Back-scattered electron image depicting the exsolution phenomenon of iron-rich orthopyroxene, (c) Clinopyroxene inclusions (remnant?) in orthopyroxene host and (d) SEM image showing fine lamellae of Cpx inclusion and host Opx. respectively) contain only secondary orthopyroxene. Sample which yields a temperature of 638 o C is also from a retrogressed quarry. The temperature dependence of Fe +2 and Mg +2 distribution between coexisting garnet and clinopyroxene has been calibrated by several workers (Dahl, 1980; Raheim and Green, 1974; Ellis and Green, 1979). Experimental calibration of Ellis and Green (1979) was applied to 14 Grt-Cpx pairs (Appendix Table 7). These temperature estimates range from 733 o C to 822 o C but most of them are between 750 o C and 775 o C. There is less scatter in these temperatures, compared to the results obtained from Grt-Opx geothermometers. From the above discussion based on the Appendix Table 7, it is evident that the temperature estimates 178 regional variation of these temperatures. GEOBAROMETRY Geobarometers involving garnet+orthopyroxene +plagioclase+quartz assemblage have been widely applied to crustal rocks (Newton, 1983; Raith et al., 1983; Schumarcher et al., 1991). There are several formulations based on either measured thermodynamic data or experimental work (Perkins and Newton, 1981; Newton and Perkins, 1982; Bohlen et al., 1983; Perkins and Chipera, 1985; Berman, 1988; Bhattacharya et al., 1991). Calibration by Newton and Perkins (1982) is based on measured thermodynamic data for the Mg end-member reaction (3) and inappropriate for the mostly iron-rich mineral compositions of

9 the present sample set (X Fe-Opx ranges from 0.45 to 0.95). Therefore, Fe end-member calibrations of Bhattacharya et al., (1991) and Perkins and Chipera (1985) were employed to calculate the palaeopressures. Pressures were also calculated using the thermodynamic data set of Holland and Powell (1998). Calculations were done at 800 o C (Appendix Table 8) as most temperature estimates are between 750 o C and 800 o C, and also considering the fact that most of the thermochemcial data have been derived in this range of temperature. However, calculation for samples B-117 and were made at 700 o C as Opx in these two samples have been formed as a result of cooling. Pressure estimates from Bhattacharya et al., (1991) are between 7 and 9.8 kb while the estimates from Perkins and Chipera (1995) are between 7.8 and 11 Kb (Appendix Table 8). For most of the samples there is a good agreement among the results obtained by different calibrations. Regional variations of the paleopressures calculated according to Perkins and Chipera (1985) is shown in Figure 5. A notable feature is that there is a pressure increase of 2-3 Kb from western parts towards the easternmost parts and Kataragama klippe, irrespective of the calibration employed. These results are compatible with the previous studies that indicated a pressure increase from west to east or southeast within the HC (Schumarcher et al., 1991) and confirms the paleopressure contouring by Faulhaber and Raith (1991). Thus, calculations at 800 o C clearly indicate that the paleo-pressures in the southeastern part of the HC were not less than 10 kb. In comparison to numerous studies based on Grt + Opx + Plag + Qtz geobarometer there is relatively fewer number of applications of Grt + Cpx + plag + qtz barometry (Perkins and Newton, 1981; Newton, 1982). Calibration by Newton and Perkins (1982) for the Cpx equilibria is based on Mg-end member reaction and therefore suitable for Mg-enriched mineral phases. Considering the ferriferous nature of the minerals encountered in the present study paleopressures were calculated at 800 o C using thermochemical data set of Holland and Powell (1998). In Appendix Table 8, results are compared with pressure estimates obtained from Grt + Opx + Pl + Qz barometer. At 800 o C, there is excellent agreement between two barometers when applied to all (12) samples of two pyroxene granulite samples encountered in our study. Geographic distribution of the resultant pressures is also very much similar to that shown by Grt + Opx + pl + qz equilibria. OPAQUE MINERALS AND OXIDIZING STATE OF CHARNOCKITIC ROCKS Ilmenite is the dominant oxide mineral that occurs with or without graphite, pyrhotite or magnetite. Magnetite is restricted to tonalitictrondhjemitic rocks which usually have relatively high Mg/Fe+Mg bulk compositions. Their ulvospinel component is less than 2% suggesting advanced re-equilibrium during cooling. Hematite component of ilmenite is also extremely low (<2%). These conclusions are based on optical examination and about 10 microprobe analyses (focussed beam-2-3 m) of opaque phases per polished section. On rare occasions exsolution of Fe-Ti oxides were observed but reintegration of original compositions were not attempted as a detailed study of opaque phases and magnetic petrology of these samples is underway. The ilmenite+hedenbergite primary assemblage and hedenbergite+ferrosolite retrograde assemblage in high X Fe charnockitic granulites implies a low oxygen fugacity with log f O2 values near or below the QFM buffer (Wones, 1989). Despite advanced re-equilibration it may be possible to estimate oxygen fugacity using pyroxene-quilf equilibria in the presence of two Fe-Ti oxides to estimate a maximum fo 2 value in the presence of ilmenite (Frost and Lindsley, 1992). Temperature calculated from two-oxide thermometry and two-pyroxene thermometry tally only for a few samples. In most cases oxide thermometry results in much lower temperatures indicating extreme resetting of cations in these minerals. Calculations yield 179

10 a range of oxygen fugacities from to bars. A notable feature is that charnockitic granitoids of granitic-adamellitic composition have the lower oxygen fugacities ( bars) while granitoids of tonalitic-trondjhemititc compositions have relatively higher fugacities ( bars). However, slightly higher oxygen fugacities might have prevailed in two types of lithologies during peak metamorphism. These observations suggest that the oxygen fugacitites of these rocks, buffered by silicate assemblage were fairly uniform throughout granitoid rocks having comparable X Fe. Magnetic susceptibility measurements and available aeromagnetic anomaly maps also indicate the presence of relatively higher amount of magnetite in tonalitic and trondjhemitic rocks. MINERAL CHEMISTRY GARNET Electron microprobe analyses of garnet cores from pyroxene granulite samples are presented in Appendix Table 2. The results show that these garnets essentially consist of almandinegrossular-pyrope solid solution. Almandine (Alm) content in granitic-ademalletic rocks is generally higher (65-74%) than that of tonalitictrodhjemitic rocks (50-71%) and restricted to a narrow range. Pyrope (Prp) content which varies from 3 to 21% displays a reverse relationship. Grossular (Grs) component of all garnets is restricted to a very narrow range of 17-24%. Rim- interior- and core areas of selected garnet grains were probed in order to ascertain any compositional zoning. In most of the garnets there is no significant zoning penetrating into the grain interior except for local zoning of rims adjacent to plagioclase and ferromagnesian minerals. However, garnet in sample 91-14, particularly smaller grains show a significant zoning in Alm and Prp contents Alm and Prp contents increase towards the interior while spasertine (Sps) and Grs increase towards the rim. This type of zoning pattern can be developed after peak metamorphism due to resorption of outer part of the garnet (Tracy, 1982). Zoning of Alm and Sps was much more significant than that of Grs and Prp because no recipient minerals of Fe and Mn were involved in resorption. ORTHOPYROXENE Enstatite and ferrosilite components of orthopyroxene from 26 samples are presented in Appendix Table 3. Positive correlation between pyroxene composition (X Fe ) and bulk compositions is illustrated in Figure 4a. Two characteristic X Fe ranges for granitic-adamellitic rocks and tonalitic-trondhjemitic rocks are also explicit in this diagram. Lower range ( ) and higher range ( ) of X Fe corresponds to tonalitic and granitic rocks respectively. Magnesian-poor granitic rocks (91-14, 90-66, , B-25) contain orthopyroxene extremely rich in ferrosilite. In fact these have been exsolved and grown from clinopyroxene as a result of cooling. Bulk compositions of rocks containing ferrosilite rich orthopyroxene have been discussed and base on their high Fe/Fe+Mg values and other chemical characteristics these rocks have been classified as A-type anorogenic granitoids (Prame, 1997). CLINOPYROXENE Clnopyroxene was present in 15 rock samples and their compositions are given in Appendix Table 4. Their wollastonite and enstatite components vary from 42 to 46% and from 4 to 34% respectively. Thus they belong to diopside-hedenbergite series. These pyroxenes are very low in Al ( p.f.u) and other non quadrilateral components. Electron microprobe analyses of certain points on lamellae of iron-rich Cpx (B-117) have given pigeonite compositions probably indicating the presence of fine Opx lamellae within these optically discernible lamellae. Dependence of X Fe Cpx on X Fe - host rock is shown in Figure 4b. 180

11 PLAGIOCLASE Plagioclase is an essential constituent in rocks of tonalitic to trondjhemitic composition. Its modal percentage is very low in granitic rocks occurring in the eastern parts (91-14, B-117). Plagioclase occurring in these rocks are low in An (14-20 %). An and Ab contents of plagioclase from 29 samples are presented in the Appendix Table 5. The variation of An content According to the nomenclature of Leake (1978) all amphiboles belong to the calcic group, irrespective of calculation method. Fe +2 /Fe +3 ratios were calculated from the electron microprobe analyses after normalizing to 13-cations (excl. Ca, Na, K). Granitic samples generally have a higher Fe +2 /Fe +3 ratios (4-12) when compared with these ratios of tonalitic samples (0.4-4). The compositions of Fig. 4 (a) Dependence of X Fe Opx on bulk FeO/FeO+MgO ratio (filled circles for tonalitie/trondhjemitie, crosses for granitic rocks), (b) Variation between X Fe -Cpx and bulk FeO/FeO+MgO ratio, (c) Variation of An content of plagioclase with CaO/CaO+Na 2 O of host rock and (d) Composition of hornblende from granitic and tonalitic rocks (symbols as for Fig. 4a) which is between 14% and 44% with X Ca rock is illustrated in Figure 4c. AMPHIBOLES Amphiboles were present in 23 rock samples (composition of amphibole formed along the edges of pyroxene are excluded). The structural formulae of apparently pre- or synkinematic amphibole analyses were calculated on13-cation (excl. Ca, K, Na) basis (Appendix Table 6). studied amphiboles are shown in relation to various end-members (Figure 4d). The positive correlation is due to exchange of edenite and tschermakite components but the deviation from end-member tie-line indicates substitution by Fe +3, Ti +4 etc. BIOTITE Biotite was present only in 10 samples. Clearly, some of them contain secondary biotite while 181

12 others exhibit ambiguous textural relationships. X Fe ranges from 0.42 to Titanium contents are generally high, the highest being 0.70 p.f.u WOLLASTONITE-SCAPOLITE-GARNET- GRANULITES AND ASSOCIATED ROCKS The following section attempts to constrain the peak metamorphic temperatures of the granulite facies metamorphism of the study-area using equilibria of calc-silicate minerals, especially in the light of available internally consistent thermochemical data, improved activitycomposition relations for grossular garnet and scapolite and well established paleopressure of the HC near the Highaland-Vijayan boundary (Figure 1). It should also be noted that there are a number of granulite facies enclaves within this Vijayan area dominantly made up of amphibolite facies rocks. Petrographic examination reveals the stable existence of following assemblages during peak metamorphism. Scapolite + diopside + wollastonite + plagioclase + calcite + grossular quartz ± vesuvianite (AJ-1) Scapolite + diopside + wolastonite + grossular + calcite + quartz (U-143) Fig. 5 Geographic variation of paleo-pressure estimates at 800 o C according to Perkins and Chipera (1985). Note the pressure increase from about 7.5 kb from western part ( i.e. Matara) to over 10 kb towards eastern part (i.e. HC/VC boundary). estimates ( 9kb) from garnet-pyroxene granulites. PETROGRPAHY As described elsewhere calc-silicate rocks mostly occur as lenses, boudins or minor intercalations with pyroxene granulites and impure marble. Samples U-143, R-163, R-206, AJ-1, and SAM-1are from the southeastern part Scapolite + diopside + plagioclase + grossular calcite quartz (R-163) Scapolite + plagioclase + calcite + diopside± graphite (R-206) Anorthite + garnet clinozoisite sphene magnetite (SAM-100) Mineral constituents of the calc-silicate rocks are given in Appendix Table

13 Fig. 6 (a) Garnet rim with quartz and calcite blebs formed between scapolite and wollastonite (U-143), (b) Garnet and quartz formed between plagioclase and wollastonite (AJ-1), (c) A garnet rim formed between scapolite and calcite, (d) Scapolite rims between vesuvianite and garnet (AJ-1), (e) Sieve garnet replacing anorthite with sporadic clinozoisite (SAM-100); note the large area in extinction probably showing a single grain of plagioclase now replaced by garnet (pseudomorph?) and (f) Clnozoisite-sieve garnet (Grs)-anorthite association in SAM-100. Though there is no significant compositional difference between porphyroblastic garnet and coronal garnet textural evidence favours the formation of prograde as well as retrograde garnet. In contrast to the charnockitic pyroxene granulites, mineral assemblages in calc granulites vary within a given layer or an outcrop. MINERAL REACTIONS As in the case of many high-grade rocks there is no direct evidence for prograde reactions in these calc-silicate rocks. However, quartz and calcite inclusions in wollastonite (sample U- 143,) are indicative of the reaction: calcite + quartz wollastonite + CO (5) RETROGRADE REACTIONS Out of the retrograde mineral reactions, those leading to the formation of coronal garnet are most prominent. These garnet forming reactions are reported from many granulite grade terrains and interpreted as a consequence of near isobaric cooling of the lower crust after peak metamorphism. In sample U-143, garnet rims with calcite and quartz blebs occur between scapolite and wollastonite (Figure 6a). This observation can be explained by the fluid-absent reaction: 183

14 scapolite + 6 wollastonite 3 garnet + calcite + 3 quartz (6) Presence of clinopyroxene (diopside) in the vicinity of some reaction domains may indicate the involvement of this mineral as a source of Fe for minor almandine component of the grossular garnet. In sample AJ-1, thin garnet and quartz rims have been formed between wollastonite and plagioclase (Figure 6b). This textural relationship is consistent with the fluid-absent reaction: 2 wollastonite + plagioclase garnet + quartz (7) Other notable textural features observed include scapolite rims formed between garnetvesuvianite (Figure 6d) and probably during a late metasomatic event. Sample SAM-100 is from a silica undersaturated boudin near the Highland-Vijayan boundary. Its bulk composition is comparable to that of an anorthosite (SiO 2 %=40, CaO%=26, Al 2 O 3% =23, FeO%=9). Over 90% this rock is made up of garnet and plagioclase. Garnet which appears to be replacing plagioclase is set in a matrix dominantly made of plagioclase, forming a sieve texture (Figure 6e). However, absence of quartz, wollastonite and scapolite makes it difficult to model garnet producing reaction that consumes plagioclase. Clinozoisite with straight contacts to garnet and plagioclase is present Fig. 7: T-X CO2 topology (a. sample R-163, b. sample AJ-1) In sample U-143, garnet rims separating scapolite from calcite (Figure 6c) can be ascribed to the decarbonation reaction: amounting to a model percentage of about 2%. There is no direct evidence to prove that this clinozoisite is a secondary product. scapolite + 5 calcite + 3 quartz 3 garnet + 6 CO (8) Other retrograde reactions indicative of subsequent cooling include rare cases of breakdown of wollastonite to form calcite + quartz (AJ-1) and decomposition of scapolite to plagioclase and calcite. A calc-silicate rock with similar textural features has been reported from the Prince Olav Coast, East Antarctica (Hiroi et al., 1987). Subramaniam (1956) has also reported somewhat similar garnet-anorthite-clinozoisite rock from Sittampundi Complex, Madras State but mineral assemblage differs in having courundum. 184

15 MINERAL CHEMISTRY OF CALC SILICATE ROCKS Representative microprobe analyses of scapolite + plagioclase and garnet + diopside from studied calc-granulite sample are given in Appendix Tables 10 and 11 respectively. GARNET Garnets are essentially grossular-andraditealmandine solid solutions with minor (usually <1%) pyrope and spessertine components. These compositions are similar to garnet compositions reported from calc silicate rocks of some other granulite terrains. Usually there is no compositional distinct between porphyroblastic garnet and coronal garnet but garnet rims between wollastonite and plagioclase (AJ-1) are highly calcic (X gr =0.92). The highest andradite content is about 13% (SAM-100). SCAPOLITE Semi-quantitative microprobe analysis scapolite from all analyzed samples indicates that they are devoid of S and amount of Cl is very low except for AJ-1. They have AnEq (= [Al-3]/3 x 100 on the basis of Si+Al=12) ) values between 55% and 78%. Scapolite compositions from sample AJ-1 vary significantly. Fresh-looking zones of scapolite are relatively low in Ca while dark and apparently altered zones are highly calcic. Variation of composition is apparently unrelated to any zoning pattern. A notable feature of the high-ca analyses is unusual Si-Al substitution resulting in high Si/Al ratio and relatively low Me ([Al-3/3]) contents. Such compositional variations within a grain and unusual Si-Al substitution is possible as a result of crystallographic constraints imposed on two independent substitution schemes (e.g. Chamberlain et al., 1985; Hassan and Buseck, 1988; Sherriff et al., 2000). PLAGIOCLASE Although An content of the different samples varies greatly, there is no marked compositional variation within an individual sample or grain. Plagioclase from SAM-100 appears to be pure anorthite (Appendix Table 10). PYROXENE Clinopyroxenes are basically diopsidehedenbergite solid solutions with variable and minor amount of Al 2 O 3 (Appendix Table 11). Mg/(Mg+Fe) ratio varies from 0.41 to T-X CO2 RELATIONS IN WOLLASTONITE- SCAPOLITE-GARNET ROCKS Electron microprobe analyses of minerals from calc-silicate rocks (Appendix Tables 10 and 11) and petrographic features indicate that garnet and diopside are low in Mg, Mn etc. and involvement of minerals such as sphene and ilmenite is also minimum. Therefore as a first approximation, mineral reactions in this rock can be modeled in CaO-Al 2 O 3 -SiO 2 - Vapour system. In present study activity model of Engi and Wersin (1987) was adopted for grossular rich garnet. Calculation of meionite activity in scapolite is critical as different models available (Moecher et al., 1991; Baker and Newton, 1995) may significantly shift the invariant points and topology. Barker and Newton (1995) model which assumes complete disorder was adopted as peak temperatures would foreseeably exceed 750 o C. Model proposed by Newton and Haselton (1982) was employed to calculate activity of anorthite. Analysis of scapolite for S and Cl from present calc-slicate rocks indicates that they are almost devoid of Cl and SO 4. Therefore in considering the fluid-mineral relations of these rocks our attention was focused on CO 2 and H 2 O. As discussed in a succeeding chapter it is reasonable to assume that peak metamorphic pressure was not less than 9kb and garnet coronas were formed at an early stage of retrograde path. Therefore, T- X CO2 diagrams were constructed for samples R- 163 (a Grs =0.7, a Me =0.3, a An =0.75) and AJ-1 (a Gr =0.85, a Me =0.15, a An =0.6), at 9kb using thermodynamic data set of Holland and Powell (1998) and PeRpLeX software by Conolly (1990). T-X CO2 topologies (Figures 7a and 7b) are similar to high-pressure (>9kb) pure CASV 185

16 topology of Warren et al. (1987) but An, Cal, and Wo invariant points for AJ-1 has slightly converged. The absence of wollastonite in R- 163 Grt+Pl+Scp+Wo assemblage with given compositions would have buffered the X CO2 around 0.25 at a temperature not less than 875 o C. For AJ-1, a T-X CO2 diagram was constructed using scapolite with highest Me content-an Eq =62. Although T-X CO2 topology is similar to that of sample R-163, it has further shrinked and shifted towards lower temperatures. Co-existence of Grt+Wo+Pl+Scp again favours a minimum temperature of 850 o C and XC O2 value of about 0.3. However, topology inverts when a Me value is about 0.2 provided that other parameters remain unchanged. As mentioned earlier, there is unusual compositional variation in scapolite from AJ-1, probably as a result of crystallographic constraints imposed on substitution reactions. On such occasions, localized ordering of Al and Si can take place vitiating the validity of an activity model based on complete disorder. Coexistence of plagioclase and wollastonite was observed only in above described samples (R- 163, AJ-1). Lack of plagioclase in U-143 of which the mineral compositions are not much different may have been caused by higher X CO2 values (>0.3) in this sample. Different reactions involved in the formation of garnet coronas in AJ-1 and U-143 also indicate evolution of their retrograde path under different X CO2 conditions (Figures 7a and 7b). Cooling is essential to form garnet by solid-solid reactions 6 and 7 while appearance of garnet rims between calcite and scapolite may have been caused by lowering of CO 2 activity. In sample R-206 collected from the same area, calcite and quartz are in direct contact but there is no wollastonite. Lack of wollastonite in this sample which could also have been heated to a temperature not less than 850 o C indicate high CO 2 activity. Presence of CO 2 bearing fluids is also indicated by intergranular graphite strings, specially between calcite and plagioclase. Thus, calc-granulites and impure marble which were metamorphosed under more or less similar P-T conditions indicate significant differences in fluid conditions. Therefore it is reasonable to assume that the any fluids affected these calcgranultes were internally buffered by mineral equilibria in individual lithologies, if not even in local domains. In fact, heterogeneity of distribution of equilibrium assemblages suggests variation of fluid composition within a given lithology. A more plausible explanation is that cooling of these rocks were caused by introduction of H 2 O which also diluted CO 2 during retrogression. THERMOMETRY BASED ON TEXTURAL RELATIONSHIPS OF CALC-SILICATE ROCKS Clear textural relationships for the fluidindependent reactions (6) and (7 ) observed in sample U-143 and AJ-1 were used to obtain univariant reaction curves using activity corrected compositions and thermodynamic data set of Holland and Powell (1998). At 9 kb, reaction (6) in sample R-163 yields a minimum temperature of 850 o C (Figure 8). Similarly, a minimum temperature of 825 o C can be estimated for reaction (7) in sample AJ-1 (Figure 8). PEAK METAMORPHIC CONDITIONS From the present geobarometric study of the garnet-pyroxene granulites as well the previous studies it can be well established that the minimum palaeo pressures in the southeastern areas of the HC are Kb (Tabl 8). These pressure estimates in conjunction with the various textural features and mineral compositions of the calc-silicate rocks can be used to ascertain the minimum peak metamorphic temperatures of the studied area. Further, scapolite with high meonite from sample AJ-1 (Appendix Table 10) also wouldn t be stable at temperatures below 850 o C. Thus it is evident that temperatures around or in excess of 875 o C prevailed during peak metamorphism. 186

17 These temperatures are comparable with the highest temperatures obtained using Grt-Px cation exchange thermometry of charnockitic pyroxene granulites.. High temperature Fig. 8 Generalized P-T diagram illustrating possible peak metamorphic conditions (stippled oval), metamorphic conditions based on present study; probably representing a reset stage and postulated retrograde path. Stippled band shows the P-T spread given by seven (07) garnet-pyroxene-plagioclase-quartz samples from the southeastern Highland Complex near the Vijayan Complex boundary. Lines 6 and 7 are for the fluid independent reactions 6 and 7 observed in calc-silicate rocks from the same area. assemblages indicating temperatures over 1000 o C are reported from elsewhere in the HC by a number of studies (Osanai et al., 2006; Dharmapriya et al., 2013; Sajeev et al., 2007). Relatively low and scattered temperature estimates from the Grt-Px thermometry may be due to efficient but varying degrees of Fe +2 Mg +2 re-equilibrium between garnet and pyroxene in different rock samples. Therefore it is reasonable to believe that temperatures yielded from calc-silicate equilibria are more representative of peak metamorphic temperatures. If this is the case, pressure estimates calculated at 800 o C (Appendix Table 8) have to be incremented by about 01 kb in order to obtain a more realistic peak metamorphic pressure for the samples from the easternmost study-area. P-T TRAJECTORY AND GEODYNAMIC IMPLICATIONS Although rare kynite inclusions in garnet have been reported (e.g. Hiroi et al., 1994; Raase and Schenk, 1984) sillimanite inclusions are relatively abundant in pelitic rocks throughout the HC including the studied area.. This indicates that the prograde path within the kyanite stability filed was short-lived and much of it evolved within sillimanite stability field (Prame,1991). In fact, Raase et al. (1994) noticed a succession of sillimanite-kayanitesillimanite inclusions distributed from core to rim. In Figure 8, stippled band shows the P-T spread given by seven (07) garnet-pyroxeneplagioclase-quartz samples from the southeastern HC near the VC boundary. Lines 6 and 7 are for the fluid independent reactions 6 and 7 observed in calc-silicate rocks from the same area. Formation of coronal garnet in calcsilicate rocks and some of the garnet pyroxene granulites can be attributed to near isobaric cooling as indicated by the arrow. Stabilization of iron-rich orthopyroxene (Fs 95 ; Appendix Table 3) instead of fayalite + quartz can also be explained by cooling under high pressure conditions. Previous studies carried out on garnet-pyroxene granulites from elsewhere in the HC have also reported textures supportive of near-isobaric cooling after peak metamorphism (Faulhaber and Raith, 1991; Schumacher et al., 1991; Prame, 1991). A previous study on a calc-silicate rock sample northeast of the study area has also given similar evidence (Mathavan and Fernando, 2001). Temperature estimates obtained using compositions of newly formed garnet and nearby clinopyroxene (Ellis and Green, 1978) are around 735 o C. Even higher temperatures ( o C) would have prevailed during the formation of garnet as our calculated temperature is likely to represent a reset stage between Grt and Cpx. Formation of orthopyroxene and plagioclase at the expense of garnet + quartz or grt+cpx+qtz in several samples (90-34 and 90-62) can be attributed to this decompression episode after near isobaric cooling. Caculations using 187